The present study consisted of a 2×2×2 design used to test the hypotheses that academic risk taking assessed with a mathematics measure would be differentially affected by goal orientation (practice goal vs test goal), competency criterion (mulit-level criterion vs minimum criterion), and evaluation source (self evaluationvs other evaluation). One hundred thirty-one college students who enrolled in an educational psychology and measurement course were systematically assigned to one of the eight experimental conditions. The Mathematics Risk Taking (MRT) measure eyielded two dependent variables: difficulty and accuracy. High difficulty and low accuracy were viewed as evidence of high risk taking. Results indicated that multi-level criterion elicited higher self assessment motive than did a minimum criterion; and self evaluation induced higher self-assessment motive than did other evaluation, whereas other evaluation promoted higher self-enhancement motive than did self evaluation. Data failed to support the proposed interaction effects between goal orientation and competency criterion on risk taking, and between evaluation source and competency criterion on risk taking. Explanations were offered for these findings and educational implications were discussed.